AI Paper Atlas

Research map / Label-efficient and robust learning / Noisy labels

Class2Simi: A Noise Reduction Perspective on Learning with Noisy Labels

Songhua Wu, Xiaobo Xia, Tongliang Liu, Bo Han, Mingming Gong, Nannan Wang, Haifeng Liu, Gang Niu

ICML 2021 · Label-efficient and robust learning › Noisy labels · approach: Noise-aware loss and pairwise methods · 20 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Transform noisy class labels into pairwise similarity labels and train with F-Class2SimiCIFAR10 · Sym-0.6accuracy: 77.03±0.92 -> 79.45±0.53Improved
Transform noisy class labels into pairwise similarity labels and train with F-Class2SimiCIFAR100 · Sym-0.6accuracy: 37.93±0.86 -> 40.38±0.58Improved
Transform noisy class labels into pairwise similarity labels and train with R-Class2SimiClothing1M* · real-world noisy labelsaccuracy: 74.98 -> 75.76Improved
Use similarity loss instead of cross-entropy on clean class labels (ablation)CIFAR100 · clean labelsaccuracy: 58.74±0.51 -> 60.36±0.89Improved
Use similarity loss instead of cross-entropy on clean class labels (ablation)MNIST · clean labelsaccuracy: 99.30±0.02 -> 99.24±0.05Hurt
Use similarity loss instead of cross-entropy on clean class labels (ablation)News20 · clean labelsaccuracy: 59.86±0.39 -> 59.74±0.20Hurt
Use similarity loss instead of cross-entropy on clean class labels (ablation)CIFAR10 · clean labelsaccuracy: 94.03±0.14 -> 94.05±0.27No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy